Impact of using a next button in a web-based health survey on time to complete and reliability of measurement Ron D. Hays (Rita Bode, Nan Rothrock, William Riley, David Cella, Richard Gershon) October 31, 2009, Presented at International Society for Quality of Life Research meeting

PROMIS Banks (454 items) Emotional Distress –Depression (28) –Anxiety (29) –Anger (29) Physical Function (124) Pain –Behavior (39) –Impact (41) Fatigue (95) Satisfaction with Participation in Discretionary Social Activities (12) Satisfaction with Participation in Social Roles (14) Sleep Disturbance (27) Wake Disturbance (16)

Study Aims Examine effects of: –Requiring (versus not) selecting Next button after responding to an item –Allowing (versus not) going back after each response to review or change it ….. on: –Time to respond to items –Number of missing responses –Reliability –Score on domain tested

Study Design Four experimental conditions 1) Automatic advance, not allowed to go back > Auto/No back 2) Automatic advance, allowed to go back > Auto/Back 3) Next after response, not allowed to go back > Next/No back 4) Next after response, allowed to go back –> Next/Back

Study Sample 807 participants in Polimetrix PollingPlace registry –Average age: 53 (range: 18-88) –64% female –87% White; 7% Hispanic; 3% African American, 3% Native American –Education Level: 2% less than HS grad; 18% HS Grad; 44% some college; 37% college+ Age, gender, race/ethnicity and education did not differ by experimental condition.

Demographics by Group Auto/NoAuto/BackNext/NoNext/Back % Female % White % Hispanic7662 % College Age (mean)565257

Sample purification Identified respondents whose response times were unreasonably fast –Deleted 18 respondents in the Auto/No back group who took an average of less than 2 seconds to answer the items To compensate for these deletions –Deleted the 18 respondents in each of the other groups with the fastest time per item 10 in Auto/Back, 2 in Next/No Back, and 1 in Next/Back groups had <2 seconds/item response

Performance of social/role activities (56 items) Items administered using 5-point frequency scale: –Never, rarely, sometimes, often, always –“I am limited in doing my work (include work at home)” –“I am able to do all of my regular family activities” –“I am able to do all of my regular leisure activities.”

Satisfaction with social/role activities (56 items) Items rated on a 5-point extent scale: –Not at all, a little bit, somewhat, quite a bit, very much –“I am happy with how much I do for my family.” –“I am satisfied with my ability to work (include work at home).” –“I am satisfied with my current level of social activity.”

Differences between groups Time to respond to items Number of missing items Internal consistency reliability Mean domain scores

Significant differences in time spent were found Automatic advance –With no back button, 13 items per minute answered –With back button, 12 items per minute When required to use the Next button –With no back button, answered 9 items per minute –With back button, 8 items per minute

Missing data and reliability did not differ by group Auto/NoAuto/BackNext/NoNext/Back Soc./Role Performance Soc./Role Satisfaction * Internal consistency reliability estimates were all * There were no significant differences in mean domain scores across groups.

Recommendations Use of automatic advance rather than Next button –Especially helpful for persons with physical limitations that could make fine motor control more difficult Use of back button –Guard against accidental key entry –Response time cost was minimal –No effect on scores or missing responses

Caveats Use of Next button has advantage of making it easier to skip items –Need “prefer not to answer” choice if using automatic advance Automatic advance doesn’t work for “select all that apply” items In CAT, allowing persons to go back to prior item is trickier

Acknowledgements PROMIS I is a U.S. National Institutes of Health (NIH) Roadmap initiative funded by cooperative agreements to a Statistical Coordinating Center (David Cella, PI, U01AR52177) and 6 primary research sites (late 2004 until 2009). PROMIS II has 12 primary research sites including –Development and Initial Validation of PROMIS GI Distress Scale –Dinesh Khanna and Brennan Spiegel, PIs

Thank you!